Improving sensing action via multi-agent based system with statistical and granular approach

With the aim to improve productivity, quality and complexity of physical processes a new class of systems called Cyber-Physical Systems (CPS) has been developed. Basically, they are made up of three principal components: a sensing part, some Decision Support Services (DSS) and an actuating part. In this work we propose a solution for a Distributed Measurement System (DMS) to be used as the sensing component of a CPS to achieve high levels of scalability, flexibility and reliability. Multi-Agent (MAS) paradigm, which resulted to be a natural approach to address the requirements of the problem, has been adopted for the DMS, in addition to statistical techniques to extract an higher level of information from the acquired data, instead of punctual values. Moreover, the concept of granularity has been applied to sensors in order to improve the robustness and the performance of the procedure. As an application of the system, we employed many Arduino® Uno micro-controller units, each one equipped with a circuit board holding I2C bus interface and a temperature sensor, to compute an indoor temperature ground truth, useful for industrial or domestic temperature monitoring.